A rapid self-healing polymer mediated by ion aggregates achieves effective encapsulation of sustainable perovskite solar cells

Abstract

Despite the rapid progress of perovskite solar cells (PSCs) toward commercialization, perovskite layers are unstable and contain toxic lead. Encapsulation represents an efficient strategy for improving the stability of PSCs as well as suppressing lead leakage. However, the damage to encapsulation materials during outdoor applications inevitably deteriorates the packaging effect and increases lead leakage. Here we report a damage perception and rapid self-healing encapsulant consisting of alkoxy polyvinylimidazole bis(trifluoromethanesulphonyl)imide (EP). The dynamic ion aggregates in the encapsulant can easily drive the molecular chain movement of EP, thereby achieving rapid damage repair to maintain device stability and inhibit lead leakage. The damaged cracks of EP completely self-heal within 6 minutes at 50°C. The EP encapsulated devices exhibit a lead sequestration efficiency of more than 99% under poor weather. After 1500 hours of damp heat test and 300 thermal cycles, the EP encapsulated devices retain 95.17 and 93.53% of their initial efficiency, respectively.

A rapid self-healing encapsulant for perovskite solar cells enhances device stability and inhibits lead leakage.

INTRODUCTION

Perovskite solar cells (PSCs) are expected to bring a revolutionary change in photovoltaics due to their remarkable photovoltaic efficiency and notably diminished cost in comparison to other alternatives. In practical applications, PSCs are inevitably damaged by natural factors such as hail, snow, wind loads, or external impacts. However, there is a response time from damage to the repair or replacement of the devices (1–3). Consequently, the infiltration of water and oxygen can easily trigger the decomposition of sensitive perovskite films, leading to device performance degradation and inducing lead leakage pollution (4–8). Therefore, it is urgently demanded to develop effective external encapsulation that can offer an inert environment functioning as a barrier against ambient oxygen and water, improve device stability, and suppress lead leakage under the premise of sustainable development (9–14).

The now conventional used encapsulants derive from silicon solar cells encapsulation technologies, including ethylene-vinyl acetate (15), polyolefin (POE) (16, 17), polyurethane (18, 19), polyisobutylene (20, 21), etc. These encapsulation materials have achieved advancements in improving device stability, but their effectiveness against lead leakage has not been extensively considered. Therefore, various encapsulants have emerged and made encouraging progress in enhancing device stability and inhibiting lead leakage (17, 22–25). For instance, an acrylate hot melt encapsulant in combination with polyvinyl butyral can prevent more than 96% of lead leakage from the devices after heavy rain for 6 hours (26). On the basis of previous research to independently address device lifetime and lead leakage issues (9, 11, 17, 21), the devices encapsulated with a polyphenol-based encapsulant can not only keep the lead content within the safe drinking water threshold but also maintain more than 90% of the initial PCE after 650 hours at 22°C and 65% relative humidity (RH) (27); meanwhile, an ultraviolet (UV) curable encapsulation material developed for monolithic perovskite/silicon tandem solar cells exhibited a lead leakage inhibition rate of ~97% after being immersed in water for 5 hours and maintained an initial efficiency of 93.5% after 1300 hours at 60°C and 85% RH (28). Furthermore, the PSCs encapsulated with a fluorosilicone polymer gel had a lead leakage inhibition rate of 99% in the rain test for 4 hours and maintained 98% of the normalized power conversion efficiency after 1000 hours at 85°C and 85% RH, satisfying the requirements of the International Electrotechnical Commission (IEC) 61215 standard (29). It is worth mentioning that Jiang et al. developed an epoxy resin with self-healing properties that was successfully used for perovskite device encapsulation. As a result, the devices encapsulated with a self-healing epoxy/UV-curable resin dual-component system can minimize the lead leakage under different weather conditions, benefiting from the ability of epoxy to repair cracks at 85°C for 1 hour (24). However, these encapsulation materials are incapable of self-repair once damaged, or the additionally introduced self-healing functional layer requires a long time to repair at high temperature, resulting in substantially reduced device stability and increased risk of lead leakage, which severely restricts their long-term application in encapsulation technology.

We proposed a rapid self-healing encapsulant mediated by ion aggregates composed of rationally designed alkoxy polyvinylimidazole bis(trifluoromethanesulphonyl)imide (EP). The EP encapsulant showed high transparency, high thermal stability, UV resistance, water oxygen barrier performance, and strong adhesion. When the EP encapsulated devices were damaged by mechanical impact in actual use, the damaged cracks of the EP encapsulant can completely self-heal within 6 min at 50°C or within 50 s at 85°C, thereby maximizing its packaging effect. The PSCs with EP encapsulation showed no distinct difference in efficiency from the unencapsulated ones, implying nondestructive encapsulation. The resulting devices retained 95.17% of their initial performance for 1500 hours under the damp heat test and maintained 93.53% of their initial performance after 300 thermal cycles, fulfilling the criteria of silicon solar cells as outlined in IEC 61215 (20). Further, even if the resultant EP encapsulated devices were damaged during the above-mentioned damp heat and cyclic thermal shocks, they can regain up to 93.1 and 87.47% of their initial efficiency through the superior self-healing performance of EP encapsulant. The EP encapsulant combined the dual characteristics of self-healing physical barrier and chemical adsorption, effectively precluding more than 99% of lead leakage in damaged devices after experiencing simulated heavy rain for 24 hours. Moreover, EP also demonstrated compatibility with flexible PSC (fPSC) encapsulation and can quickly repair bending cracks caused by repeated folding of devices.

RESULTS

Synthetic strategy and physicochemical properties of EP

We constructed a cationic polymer encapsulant EP from alkoxyimidazole through alkylation, ion exchange reactions, and free radical polymerization (Fig. 1A; see Materials and Methods for details). Typical POE was used as reference encapsulant in this work (see reason details in note S1) (30, 31). The molecular characteristics indicated by ¹H and ¹³C nuclear magnetic resonance (NMR) spectroscopy, matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry, and Fourier transform infrared spectroscopy (FTIR) confirmed the successful synthesis of the EP encapsulant (figs. S1 to S6). The molecular weights of the EP polymer is ascertained by gel permeation chromatography (GPC; fig. S7). In Fig. 1B, the EP presented good thermal stability with a decomposition temperature of up to 361.4°C. For most ionic polymers, the high glass transition temperature (T_g) causes them to exist in the form of solid powders at room temperature and without adhesive properties (32–34). In view of this, the low T_g of the ionic polymer EP was reduced to 5.06° by introducing a flexible alkoxy side chain (Fig. 1C), which laid the foundation for forming a strong adhesive encapsulation material, and a melting temperature (T_m) appeared at 98.26°C. The adhesion strength of EP was measured by tensile shear test (Fig. 1D and fig. S8), with an adhesion strength of 4.15 MPa and a fracture strain of 15.9%, having high adhesion strength and appropriate toughness (fig. S9 and table S1) (11, 25, 26, 35). Even after soaking in water for 1 week, it still maintained a high adhesion strength of 3.65 MPa, and the adhesion strength recovered to 4.11 MPa when the glass was rebonded after the EP encapsulant was dried, with a recovery rate of 95% (fig. S10). Moreover, the FTIR spectrum of the dried EP encapsulant (removed from the glass and vacuum-dried) showed no new peaks or peak shifts (fig. S11), suggesting that the microstructure of EP has not changed.

Fig. 1. Synthetic strategy and physicochemical properties of EP encapsulant for PSCs.

(A) Synthesis of EP encapsulant. (B) Thermogravimetric analysis (TGA) and (C) DSC curve of EP encapsulant. (D) The adhesion strength of EP on glass determined by tensile test. (E) Simulated conformation of EP on glass substrate and corresponding hydrogen bond interactions at the interface between EP encapsulant and glass, while (HO)₃SiOSi(OH)₄OSi(OH)₃ serves as a model for glass and R represents alkyl chain. The unit of distance between atoms is angstrom. (F) Water contact angles of EP encapsulant. (G) Transmittance spectra of glass with and without EP encapsulant (thickness, 600 μm). Inset: EP film image with an area of 15 mm by 15 mm on glass substrate. (H) Stability of EP encapsulant under UV illumination. a.u., arbitrary units.

Molecular dynamics simulation was implemented to explore the interaction between EP and glass substrates (36). The results showed that the Si─(Si)O⋯H─Im and Si─O─H⋯O(alkoxy) hydrogen bonds were formed at the EP/glass interface (Fig. 1E, Im represents imidazole), which could enhance the buffering effect of the cover glass. It should be noted here that compared with POE encapsulant that is completely composed of C─H bonds, the C(2)─H bond in the imidazole ring of EP exists in a unique chemical environment. It is located between two nitrogen atoms. As shown in fig. S12, this structure causes the electron cloud of the C(2)─H bond to be highly polarized toward the nitrogen atoms, endowing the hydrogen atom with high positivity. As a result, it can strongly attract oxygen atoms in glass to form hydrogen bonds [Si─(Si)O⋯H─Im] (34). Therefore, EP achieves strong interfacial adhesion through the Si─(Si)O⋯H─Im hydrogen bond, which is fundamentally different from the conventional C─H alkyl bonds (such as those in POE). The TFSI⁻ anion, with its large steric volume and highly delocalized charge (discussed later), can prevent the direct formation of hydrogen bonds or chemical bonds with the glass surface. Its hydrophobicity helps exclude interfacial water molecules, thereby enhancing interfacial contact between EP and glass, particularly in humid environments. In addition, as a weakly coordinating anion, TFSI⁻ reduces electrostatic interactions with imidazolium cations, promoting their attraction to the glass surface.

The water contact angle of EP film was 87.1° and remained stable at 85.2° after 10 min (Fig. 1F). Moreover, the calculated surface energy of EP (31.8 mN/m; fig. S13) features a high dispersion component (γ^d_s = 26.5 mN/m) that matches well with glass (37), enabling initial adhesion, while hydrogen bonds further strengthen the interfacial bonding. The low polarity component (γ^p_s = 5.25 mN/m) indicates that the EP has a low affinity for water molecules, which, coupled with its low water vapor transmission rate (WVTR; fig. S14 and table S2), demonstrates that the EP can provide effective moisture protection for encapsulated PSCs (26, 38). The transmittance of EP deposited on glass exceeded 90% from 400 to 900 nm (Fig. 1G), suggesting that the efficiency of devices in capturing visible light was not compromised by the encapsulation of EP/glass. The EP film was also exposed to UV radiation with a wavelength range of 280 to 400 nm and an intensity of 120 W/m² for 700 hours (Fig. 1H), and its transmittance decreased by less than 1% after a total irradiation of 84 kWh/m² (IEC62108 UV test) (17), showing good UV stability. Moreover, the adhesive strength of EP still maintains 4.13 MPa after UV radiation for 700 hours, demonstrating reliable adhesive durability (fig. S15). The temperature change tracking test results indicated that EP had a better heat dissipation capability than POE (fig. S16).

Self-healing properties of EP encapsulant

The self-healing performance of EP encapsulant under sunlight-induced heating was investigated (Fig. 2A). The PSCs adopted an encapsulation structure that combined encapsulant with the front and rear cover glasses (fig. S17) (10, 29, 38). More robust fluorescence emission peak intensity as well as a stable light absorption can be observed in both the perovskite films before and after EP encapsulation, indicating high compatibility between EP encapsulant and perovskite films (figs. S18 and S19). A metal ball was dropped from a constant height on top of the encapsulated device to simulate external impact and mechanical damage (FM 44787 testing standard) (39). The control device was crushed into several fragments, and the POE encapsulated device also presented severe damage (fig. S20). Conversely, the EP encapsulated device showed typical star-shaped cracks at the impact site, with the back cover glass remaining intact, potentially due to the dense hydrogen bonding networks strongly inhibiting crack propagation along the interface, concentrating the impact energy at the impact point. Meanwhile, the flexible alkoxy side chains absorb energy through molecular chain extension and conformational changes, prompting stress to be released in the form of local star-shaped cracks. In contrast, POE (C─H bonds with lower electronegativity difference) cannot establish effective interfacial bonding due to the lack of strong polar bonds. It only relies on weak van der Waals forces, and this causes interface debonding and overall collapse of the encapsulated device when affected. A through-scratch created by a knife was introduced on the EP encapsulant at the cover glass cracks to adequately assess its repair capability (fig. S21). The scratches of the EP encapsulant can be completely self-healed within 6 min at 50°C (normal working temperature of solar module; Fig. 2B) and repaired at 85°C in merely 50 s. Whereas the cracks of the POE encapsulant remained unchanged under different temperatures (Fig. 2C). Note that solar cells may become hotter and exceed 50°C during normal operation.

Fig. 2. The self-healing properties of EP encapsulant.

(A) Schematic diagram of self-healing of EP encapsulated PSCs after damage in practical applications, including heavy rain, hail, wind loads, or falling objects. The self-healing process of (B) EP and (C) POE encapsulants in damaged devices over time at different temperatures (50°C, temperature of the solar cells during normal operation). These optical images were observed at the glass breaks.

A self-healing mechanism of the EP encapsulant was proposed. The high-resolution transmission electron microscopy (HRTEM) images displayed the formation of relatively homogeneous spherical ion aggregates in the EP encapsulant (Fig. 3, A and B). According to the energy-dispersive spectroscopy (EDS) (Fig. 3B), it can be seen that N, O, F, and S elements (characteristic elements in EP) exist in ion aggregates, indicating that the ion aggregates are formed by the aggregation of anions and cations in EP polymers through strong electrostatic interactions, and these aggregates are dispersed in the polymer matrix (Fig. 3C). The Fourier transform of the four boxes in Fig. 3D witnesses how the microstructure of EP polymer changes from box 1 to box 4, that is, the structural change from the amorphous nonaggregate to the crystalline ion aggregates. Just as shown in Fig. 3E, this is the structural change of the crystalline-amorphous interface in the edge region of the aggregates at area I in Fig. 3D. The 0.3-nm lattice stripes observed in the HRTEM image are most likely to correspond to the lattice distance between pendant groups, which is consistent with the WAXS peak at 3.15 Å. These results indicate that the crystalline regions found in the HRTEM only exist within ion aggregates regions, and the regular crystalline packing structures of ion aggregates have been observed.

Fig. 3. Validating the self-healing mechanism of EP encapsulant.

(A) HRTEM images of ion aggregates in EP encapsulant. (B) The localized enlarged image of (A) and the EDS element mapping images of N, O, F, and S distribution in ion aggregates. (C) Schematic diagram of the composition of ion aggregates by the assembly of anions and cations in EP encapsulant. (D) The localized enlarged image of (B) and the Fourier transforms of the image involving the four box areas are displayed. (E) The enlarged image of area I in (D), showing local periodic lattice in agreement with the Fourier transform patterns. (F) The variable temperature WAXS of EP encapsulant and schematic diagram of the corresponding distances d₁, d₂, and d₃ according to the WAXS profiles. (G) Scheme of the self-healing performance driven by ion aggregates in EP encapsulant. (H) The calculated electrostatic potential of TFSI⁻ and the typical distance between two main chains marked in simulated structures of EP encapsulant.

The variable temperature wide-angle x-ray scattering (WAXS) measurement was used to gain insights into the internal structure of the EP encapsulant (40). As depicted in Fig. 3F, the WAXS spectra exhibited three peaks q₁, q₂, and q₃ within the scattering vector q = 2.5 to 25 nm⁻¹. According to the Bragg formula (d = 2π/q), the corresponding d₁, d₂, and d₃ were 15.65, 6.45, and 3.15 Å at 25°C (41, 42), representing the distances from the backbone to backbone, anion to anion, and pendant to pendant groups of the EP encapsulant (43, 44), respectively. With increasing temperature, q₂ and q₃ decreased, which was due to the relatively smaller volume of anions and pendant groups compared to the long backbone structure of the EP, enabling their enhanced mobility. By analyzing the WAXS scattering signals via peak splitting (fig. S22) (45, 46), it was found that due to enhanced ion dynamic recombination and chain segment migration, the increase in temperature promotes the proportion of crystallization in the EP encapsulant to first increase and then decrease, thereby greatly improving the self-healing efficiency and achieving rapid repair at high temperatures. On the basis of this, it can be inferred that the T_g of EP might be associated with the segmental motion of amorphous region, while the crystalline areas require a higher temperature to rearrange.

Combining TEM and WAXS results, it can be concluded that the amorphous region of EP is in a highly elastic state at room temperature, with enhanced chain segment mobility, thereby endowing the EP encapsulant with self-healing ability. At the same time, the crystalline regions form a rigid interconnected networks with tightly packed molecular chains, which avoids overall softening of the polymer. This structural arrangement not only enhances the mechanical strength of the EP encapsulant but also effectively suppresses crack propagation, ultimately creating a unique “soft-rigid” synergistic structure. The WAXS data reveal that these crystalline ion aggregates are not permanent but exhibit a temperature-dependent dynamic reversible behavior. The change in the proportion of crystalline regions indicates that these aggregates of EP encapsulant undergo dissociation, migration, and recombination. This process directly drives self-healing, that is, the increase in temperature weakens the electrostatic binding force between cations and anions in EP polymer, causing ions to dissociate from their original aggregates and migrate toward the crack interface to provide charge compensation and form new ion pairs or aggregates (Fig. 3G).

Thermodynamically, self-healing is a process driven by enthalpy and promoted by entropy increase. Differential scanning calorimetry (DSC) characterization confirms that the recombination of ions into aggregates is exothermic (ΔH < 0), and the released bonding energy compensates for the entropy of dissociation and migration, thus providing the core driving force. Actually, during the propagation of the crack, new surfaces of the materials gradually appear, which reduces the entropy, and meanwhile many ion aggregates at the material surface are destroyed, which could lead to additional entropy loss. Therefore, the self-healing could somehow be both enthalpy and entropy favorable. Simultaneously, the increase in temperature not only weakens the electrostatic interaction to trigger ion dissociation but also enhances the segment motion (increasing entropy, ΔS > 0), thereby providing conditions for migration. The overall spontaneity of this process is determined by the negative change in Gibbs free energy (ΔH − TΔS = ΔG < 0). Meanwhile, the polymer segments covalently connected to the ionic groups diffuse toward the crack area under the driving force of ion migration, filling the crack gaps. Therefore, the self-healing behavior of EP polymer stems from the synergistic effect of multiple intermolecular forces, driven by the thermally triggered reversible evolution of the ion aggregates.

In addition, the large volume and highly delocalized charge of the TFSI⁻ anion in EP made its electrostatic interaction with imidazole cations on monomer units in the same chain relatively weak (Fig. 3H) (47, 48), resulting in the tightly packed structure of polymer backbones, which also enabled the EP encapsulant to achieve a balance between self-healing and mechanical strength (49). Consequently, an ion polymer EP encapsulant for PSCs with rapid self-healing ability was obtained on the basis of the driving function of ion aggregates. This would allow self-repairing of cracks when the encapsulant undergoes external impacts, maintaining the mechanical integrity and preventing lead leakage of devices.

Lead leakage tests

Apart from the self-healing property, the electron-rich alkoxy groups contained in EP can also strongly capture the leaked lead, as confirmed by the chemical interaction between EP and lead (figs. S12, S23, and S24) and the study on adsorption kinetics (fig. S25). Therefore, the dual characteristics of self-healing physical barrier and chemical adsorption of the EP encapsulant could effectively reduce lead leakage (Fig. 4A). We simulated different real-life weather conditions to comprehensively quantify the leaked lead of EP encapsulated devices after damage, such as continuous heavy rain, acid rain, or rain and shine after hail. Among them, a falling metal ball was used to simulate extreme hail impact (according to the FM 44787 standard) to mechanically break the encapsulated devices, and a homemade device was used to simulate a heavy downpour after hail. The contaminated rainwater was sampled at different times, and inductively coupled plasma mass spectrometry (ICP-MS) was used to measure the leaked lead amounts of the damaged devices (fig. S26). As shown in Fig. 4B, the lead leakage amounts of the control and POE encapsulated devices were as high as 140.25 ± 16.71 and 65.56 ± 8.48 parts per million (ppm) after continuous rainfall for 24 hours, which decreased to 0.16 ± 0.02 ppm for the EP encapsulated devices, and the lead sequestration efficiency (SQE) was 99.89%. Under severe weather conditions, when acid rain dripped on the damaged devices for 24 hours, the lead leakage amounts of the control, POE, and EP encapsulated devices were 395.6 ± 26.48, 199.0 ± 27.29, and 4.87 ± 1.10 ppm (fig. S27), respectively, and the lead SQE of EP was 98.76%.

Fig. 4. Lead leakage of the encapsulated PSCs.

(A) Illustration of inhibiting lead leakage via the dual characteristics of self-healing physical barrier and chemical adsorption of EP encapsulant. Lead leakage test: (B) continuous normal rainy for 24 hours; (C) first exposed to sunshine for 10 min, followed by 24 hours of rainfall; (D) it rained for the first 6 hours, followed by sunny for 10 min and then raining again for 18 hours. (E) Neutral water soaking measurements for damaged PSCs with and without encapsulation.

Moreover, we heated these damaged devices at 50°C to simulate a reasonable temperature on a sunny day. As shown in Fig. 4C, when the sun shone for 10 min and then it rained for 24 hours, the differences in lead leakage were substantial. The control and POE encapsulated devices showed lead leakage amounts of 131.82 ± 5.15 and 68.82 ± 6.48 ppm, while the leakage of the EP encapsulated devices was as low as 0.021 ± 0.004 ppm. If it rained for the first 6 hours, followed by sunshine duration of 10 min, and then continued to rain for 18 hours (Fig. 4D), the lead leakage amount decreased from 84.96 ± 7.08 ppm (POE) to 0.09 ± 0.01 ppm (EP). The slight increase in temperature caused by sunlight exposure could result in a more pronounced reduction in the lead leakage of PSCs encapsulated by EP compared to other encapsulants, and the SQE was as high as 99.95%. These results suggested that the sunshine heating enabled the damaged EP encapsulant to self-repair, thereby leading to differences in lead leakage amounts between devices encapsulated with EP and POE.

Furthermore, we also conducted water-soaking tests on damaged devices to simulate an extreme scenario (fig. S28). As displayed in Fig. 4E, the leaked lead contents were relatively low within the first 2 hours, which might be due to the water gradually infiltrating from the cracks in the encapsulation layer into the perovskite layer. As the water absorbed by perovskite film increased, the control and POE encapsulated devices exhibited quick lead escape. In sharp contrast, the EP encapsulated device achieved a 98.9% SQE, which might be due to the hydrogen bonding between the EP encapsulant and the cover glass enhancing the buffering effect of cover glass, reducing the occurrence of destructive cracks in devices, and effectively reducing water infiltration. Acid water (pH = 4.2) was also used to soak damaged devices (fig. S29), and the control and POE encapsulated devices exhibited more severe lead leakage, while the lead concentration of the EP encapsulated devices was suppressed, maintaining a SQE of 94.5%.

Photovoltaic performance

The structure of indium tin oxide (ITO)/NiO_x/CsMAFA-based perovskite [(methylammonium (MA) and formamidinium(FA)]/PC₆₁BM ([6,6]-phenylC₆₁-butyric acid methyl ester) + C₆₀/BCP (bathocuproine)/Au inverted PSCs were fabricated, and the typical current-voltage (J-V) curves for unencapsulated, POE, and EP encapsulated devices were measured (Fig. 5A). The control device had a PCE of 22.36%, with short-circuit current density (J_SC) of 25.25 mA/cm², open-circuit voltage (V_OC) of 1.12 V, and fill factor (FF) of 78.85%. There was almost no difference in device efficiency after EP encapsulation (22.07%), with a V_OC of 1.12 V, a J_SC of 25.12 mA/cm², and an FF of 78.67%. By comparison, the POE encapsulation slightly declined the PCE (21.82%) with a J_SC of 24.98 mA/cm², a V_OC of 1.11 V, and an FF of 78.59%, which could be due to the heating and lamination process of POE. The photovoltaic parameter statistic distribution of these devices demonstrated the reproducibility of the high compatibility encapsulation strategy (fig. S30).

Fig. 5. Photovoltaic performance and stability of the encapsulated PSCs.

(A) J-V curves of devices before and after encapsulation. Normalized PCE variation of devices with EP encapsulation under (B) damp heat test at 85°C and 85% RH, and (C) thermal cycling test from −40° to 85°C. Error bars in (B) and (C) represent the SDs from the statistical results of different devices. Epoxy resin was used as an edge encapsulant around the device for stability test.

In accordance with the IEC 61215 standard (50, 51), the encapsulated devices were subjected to damp heat test (85% RH, 85°C) for more than 1500 hours (Fig. 5B). The efficiency of the control devices decayed by more than 50% after about 50 hours under the double infliction of moisture and high temperature, while the devices encapsulated with POE maintained 88.07% of their initial PCE after 1500 hours. In contrast, the devices with EP encapsulation exhibited satisfactory damp heat stability, retaining 95.17% of their initial PCE after 1500 hours. The cover glass itself has micrometer-scale irregularities and stacking gaps and only provides physical support, while the EP encapsulant achieves molecular-level sealing by filling all microgaps and forming the hydrogen-bonded interface to block the channels of water and oxygen. Effective long-term device encapsulation relies on the synergistic effect between the mechanical strength of glass and the active barrier function of EP. To simultaneously verify the impact of EP self-healing performance on device stability, the intact encapsulated devices were first subjected to damp heat test for 500 hours, and then the edge encapsulant of the packaged devices was artificially damaged and the efficiency changes of the encapsulated devices were continued to be recorded (details were given in device characterization). The results showed that the fast self-healing ability of EP at 85°C enabled the damaged EP encapsulated devices to still maintain 93.1% of their initial efficiency after 1000 hours of subsequent damp heat test, with minimal loss of efficiency. In contrast, the efficiency of the damaged POE encapsulated devices declined by about 50% for 744 hours under the subsequent damp heat test condition. Because of a penetrating skeleton of interconnected network and residual crystals, EP still retains elastic solid properties (no flow) at 85°C, thus maintaining a dense barrier layer.

Furthermore, we also investigated the stability of encapsulated devices under cyclic thermal cooling. The results showed that the control devices markedly decayed to 25.57% of their initial PCE after 40 cycles, in contrast to that of 93.53% for EP encapsulated devices after 300 cycles (Fig. 5C), exceeding the IEC 61215 requirement of a 5% drop after 200 cycles (95.43% after 200 cycles). Simultaneously, the edge encapsulant of the intact encapsulated devices was artificially damaged after 50 cycles of thermal shocks, and the efficiency changes of the encapsulated devices were continuously recorded. The efficiency of the damaged POE encapsulated devices deteriorated rapidly under cyclic thermal shocks, while the damaged EP encapsulated devices showed a slight efficiency decline, maintaining 87.47% of their initial efficiency after the subsequent 250 cycles. Furthermore, the operational stability of EP encapsulated PSCs retained 86.7% of their initial efficiency after 1200 hours, compared with 78.6% for POE encapsulated devices (fig. S31). The EP was also used to encapsulate devices with various perovskite compositions and structures (figs. S32 to S35 and tables S3 to S5; including fluorine-doped tin oxide (FTO)/NiO_x/CsPbI₂Br/TiO₂/Au, ITO/TiO₂/MAPbI₃/Spiro-OMeTAD/Au, and perovskite module; the comparison of device performances before and after encapsulation as well as the stability tests validated the universality of EP in PSC encapsulation). Meanwhile, these devices encapsulated with EP demonstrated an excellent lead leakage suppression effect in the drip test (fig. 36). These results convincingly proved that the EP encapsulant has comprehensive advantages of good operability, greatly enhancing device stability and suppressing lead leakage, showing great potential for application in the commercial market.

Performance of EP encapsulant on flexible devices

We also evaluated the compatibility of the EP encapsulation strategy with fPSCs. The commercially available POE is relatively thick, with a thickness of 200 to 400 μm, which limits the bending of fPSCs encapsulated with POE (16). The bending diameter of EP encapsulated fPSCs reached 3 mm (Fig. 6A). However, the encapsulant is prone to creasing after frequent bending of fPSCs, which is not conducive to maintaining long-term performance stability of the devices. The EP encapsulated fPSCs were subjected to 5000 bending tests with a radius of 3 mm, and the results showed that the bending creases of EP were completely healed at 50°C within 5 min and only took 40 s to heal at 85°C (Fig. 6B). The fPSCs showed a PCE of 19.17% with a J_SC of 23.54 mA/cm², a V_OC of 1.08 V, and an FF of 75.28%. The EP and POE encapsulated fPSCs exhibited PCE of 17.73 and 17.64%, respectively (Fig. 6C and table S6), which were similar to that of unencapsulated devices. The statistical distribution of photovoltaic parameters of these flexible devices proved the repeatability of the highly compatible packaging strategy (fig. S37).

Fig. 6. Performance of EP encapsulant on flexible devices.

(A) Schematic diagram of self-healing after damage in practical applications of flexible devices encapsulated with EP. The picture illustrates the bending diameter of EP encapsulated fPSCs reached 3 mm. (B) Optical micrographs of EP encapsulant self-healing process at different temperatures after 5000 bending cycles with a radius of 3 mm for the encapsulated flexible device. (C) J-V curves of fPSCs before and after EP encapsulation. (D) Normalized PCE variation of flexible devices with EP encapsulation after 1000 bending cycle tests and then under the conditions of 50°C and 40 ± 5% RH.

Furthermore, the encapsulated devices were tested for 1000 bending cycles with a radius of curvature of 3 mm, and then they were placed in ambient air at 50°C and 40 ± 5% RH for 1500 hours to assess the self-healing ability of the encapsulant EP in the operation of fPSCs (Fig. 6D). The POE and EP encapsulated devices was similar during continuous bending stage, retaining 82.13 and 85.23% of their initial PCE after 1000 bending cycles, respectively. The rapid self-healing ability of EP enabled its encapsulated flexible devices to effectively block moisture and oxygen erosion and maintain 80.17% of their initial efficiency at 50°C and 40 ± 5% RH, while the POE encapsulated devices decreased to 70.43% of their initial performance. During this process, 1000 bending cycles (minutes in duration) induce micrometer-scale cracks in the encapsulation layer. The encapsulants mainly act as elastic buffers for a short time, resulting in comparable efficiency retention for devices encapsulated with POE and EP. In subsequent damp heat test, these cracks then become initial channels for environmental erosion. The POE encapsulated devices exhibit crack propagation and accelerated efficiency decay due to the lack of self-healing ability, while the efficiency of EP encapsulated devices tends to stabilize (blue arrow in Fig. 6D). These results indicated that the self-healing characteristics mediated by ion aggregates of the EP encapsulant can dynamically repair the bending cracks of flexible devices, thereby extending their service life.

DISCUSSION

In summary, we successfully designed a rapid self-healing EP encapsulant for PSCs and investigated the self-healing mechanism mediated by ion aggregates as well as its effect on the stability and lead leakage of encapsulated devices. The damage cracks of EP encapsulant can be self-repaired within 6 min at 50°C and can be healed within only 50 s at 85°C. The EP encapsulated devices retained 95.17% of their initial PCE for 1500 hours under the damp heat test and 93.53% of their initial PCE after 300 cycles in the thermal cycling test. Even after the EP encapsulant was damaged in these stability tests, the encapsulated devices can still maintain good efficiency. Further, the EP encapsulated devices had a high lead SQE of more than 99% under rainy days in different scenarios and more than 98% in water soaking test through the dual effects of self-healing physical barriers and chemical adsorption. This work also demonstrated the feasibility of the EP encapsulant for devices with diverse structures, showing great promise in the development of the device encapsulation and laying a foundation for promoting the transition of PSCs to sustainable photovoltaic technology. Our future work will focus on developing environmentally adaptive intelligent encapsulation materials, such as those with mechanical stress-triggered self-healing property, to enhance the device stability in extreme environments.

MATERIALS AND METHODS

Materials

N-vinylimidazole, 1-(2-(2-methoxyethoxy)ethyl) bromide (C₅H₁₁BrO₂, 98%), and lithium bis(trifluoromethanesulphonyl)imide (LiTFSI, 99%) were obtained from Alfa Aesar. Nickel (II) nitrate hexahydrate [Ni(NO₃)₂·6H₂O, 99.99%], sodium hydroxide (99.99%), titanium tetrachloride (99.9%), ethanol (99.9%), isopropanol (IPA; 99.9%), dimethyl sulfoxide (DMSO; 99.8%), N,N-dimethylformamide (DMF; 99.8%), chlorobenzene (99.8%), and 2, 2′-azobis(2-methylpropionitrile) (AIBN; 98%) were purchased from Sigma-Aldrich. The raw materials for preparing PSCs were purchased from Xi’an P-OLED Corp. All above materials were used without further purification. The reference encapsulant of POE was purchased from First Applied Material Co. Ltd. The synthesis of NiO_x and TiO₂ refers to our previous works (52, 53).

Synthesis of 1-(2-(2-methoxyethoxy)ethyl)-3-vinylimidazolium bromine

C₅H₁₁BrO₂ and N-vinylimidazole (molar ratio is 1.5) were dissolved in DMSO solvent in a flask, which were stirred at 90°C for 48 hours under the protection of nitrogen. After cooling to room temperature, ethyl acetate solvent was added dropwise to the mixture, and the mix solution precipitated and separated gradually. The underlying crude product was dissolved in water, extracted three times with ethyl acetate, and then dried under vacuum at 60°C for 24 hours to obtain the 1-(2-(2-methoxyethoxy)ethyl)-3-vinylimidazolium bromine.

Synthesis of 1-(2-(2-methoxyethoxy)ethyl)-3-vinylimidazolium bis(trifluoromethanesulphonyl) imide

The aqueous solution of LiTFSI salt was added dropwise to 1-(2-(2-methoxyethoxy)ethyl)-3-vinylimidazolium bromine at a molar ratio of 1.2. Subsequently, the mixture was stirred at room temperature for 12 hours. The precipitate was filtered, washed several times with deionized water, and dried under vacuum at 65°C for 24 hours to obtain 1-(2-(2-methoxyethoxy)ethyl)-3-vinylimidazolium bis(trifluoromethanesulphonyl)imide.

Synthesis of polymerization 1-(2-(2-methoxyethoxy)ethyl)-3-vinylimidazolium bis(trifluoromethanesulphonyl)imide

1-(2-(2-Methoxyethoxy)ethyl)-3-vinylimidazolium bis(trifluoromethanesulphonyl)imide monomer was dissolved in DMSO solution in a flask, stirred, and used AIBN as an initiator in a nitrogen atmosphere to initiate free radical polymerization reaction at 80°C for 12 hours, and the mass ratio of initiator to monomer is 1%. After cooling to room temperature, the mixture was added to excessive ethyl acetate to precipitate the crude product of this polymer and washed several times with ethyl acetate until the washing solution was not turbid. Then, the precipitate was dried under vacuum to obtain the target polymerization 1-(2-(2-methoxyethoxy)ethyl)-3-vinylimidazolium bis(trifluoromethanesulphonyl)imide (termed as EP).

Fabrication PSCs based on MAPbI₃ absorbing layer

The ITO glasses (1.5 cm by 1.5 cm) were washed sequentially with deionized water, acetone, and ethanol, followed by UV ozone plasma treatment for 15 min. The TiO₂ precursor solution was spin-coated on ITO substrates at 4000 rpm for 50 s and heated at 160°C for 10 min. The MAPbI₃ precursor solution containing MAI, PbI₂ (molar ratio = 1:1), and solvent γ-butyrolactone/DMSO (volume ratio = 7:3) was spin-coated onto the TiO₂ layer at 3000 rpm for 50 s. Chlorobenzene solvent (180 μl) was added in the last 35 s of this procedure. The MAPbI₃ films were annealed at 100°C for 10 min. The precursor solution of hole transport layer was composed of 2,2′,7,7′-tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9′spirobifluorene (spiro-OMeTAD; 75 mg), 4-tBP (30 μl), and 4-tert-butylpyridine (4-tBP);Bis(trifluoromethane)sulfonimide lithium salt (Li-TFSI; 520 mg in 1 ml of acetonitrile) dissolved in chlorobenzene (1 ml). Subsequently, this spiro-OMeTAD mixed solution was spin-coated on perovskite film at 4000 rpm for 40 s and then kept in a desiccator for 12 hours. Last, a thermal evaporation method was adopted to successively evaporate Au electrodes (100 nm) at 2 × 10⁻⁶ mbar. The effective area of the device was 0.1 cm².

Fabrication PSCs based on CsPbI₂Br absorbing layer

The FTO glasses (1.5 cm by 1.5 cm) were washed with deionized water, acetone, and ethanol sequentially, followed by UV ozone plasma treatment for 15 min. The NiO_x precursor solution was spin-coated on the cleaned FTO substrates at 4000 rpm for 30 s and then heated at 100°C for 10 min. Afterward, the CsPbI₂Br precursor solution containing CsI, PbBr₂, PbI₂ (molar ratio = 2:1:1), and solvent DMSO/DMF (volume ratio = 1:4) was deposited on NiO_x film at 1000 rpm for 15 s and 4000 rpm for 45 s. Chlorobenzene solvent (140 μl) was added in the last 20 s of this procedure. The CsPbI₂Br films were annealed at 50°C for 2 min and at 160°C for 10 min. After that, the TiO₂ precursor solution was spin-coated on perovskite film at 3000 rpm for 50 s and then heated at 160°C for 10 min. Last, a thermal evaporation method was adopted to successively evaporate Al electrodes (100 nm) at 2 × 10⁻⁶ mbar. The effective area of the device was 0.1 cm².

Fabrication PSCs based on CsMAFA absorbing layer

The glass/ITO and polyethylene terephthalate (PET)/ITO substrates were sequentially washed in an ultrasonic bath of water, acetone, and ethanol and then treated by UV ozone for 15 min. The NiO_x precursor solution (20 mg/ml) was spin-coated on the substrates at 3000 rpm for 40 s and heated at 100°C for 10 min. The perovskite precursor solution containing formamidinium iodide (FAI) of 1.19 M, MABr of 0.21 M, CsI of 1 M, PbI₂ of 1.30 M, PbBr₂ of 0.21 M, and poly(1-vinyl-3-ethyl-acetate) imidazole tetrafluoroborate as additives in the mixture solvent of DMF and DMSO (volume ratio = 4:1) was stirred for 6 hours and spin-coated on NiO_x film at 1000 rpm for 10 s and at 6000 rpm for 30 s. Chlorobenzene (110 μl) was spin-coated on the substrate of this procedure within the last 20 s and annealed at 100°C for 20 min. Then, the mixture solution of PC₆₁BM (20 mg/ml) and C₆₀ (5 mg/ml) in chlorobenzene was spin-coated at 5000 rpm for 40 s and annealed at 60°C for 10 min. The BCP solution (0.5 mg/ml in IPA) was spin-coated at 5000 rpm for 40 s and annealed at 60°C for 20 min. After that, the Cr (4 nm) and Au (100 nm) were sequentially deposited by a vacuum evaporation equipment at 4 × 10⁻⁶ mbar. For device stability, the preparation process of the electron transport layer and barrier layer was changed from spin coating to vacuum evaporation. That is, C₆₀ (25 nm) was deposited on the surface of the perovskite thin film at an evaporation rate of 0.2 Å/s, and then BCP (8 nm) was deposited at a rate of 0.5 Å/s (pressure below 5 × 10⁻⁴ Pa). The effective area of the device was 0.1 cm².

Fabrication of perovskite solar modules

FTO glasses were prepatterned by laser etching technology, followed by ultrasonic cleaning in deionized water, acetone, and ethanol for 20 min, and then treated with UV ozone for 20 min. The poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA) film was deposited by spin coating at 5000 rpm for 30 s and then annealed at 100°C for 10 min. Subsequently, the perovskite precursor solution was blade-coated on the PTAA substrate with a spacing of 100 μm at a movement speed of 20 mm/s. Meanwhile, 20 psi (0.14 MPa) of N₂ gas was blown onto the substrates during the blade-coating process. Afterward, the perovskite film was annealed at 120°C for 20 min. Last, the C₆₀ (25 nm), BCP (4 nm), and Cu (100 nm) were sequentially deposited by a vacuum evaporation equipment at 4 × 10⁻⁶ mbar.

Device encapsulation

EP blanket encapsulation

The EP encapsulant heated to 100°C was applied uniformly to a cover glass, and then the cover glass with EP encapsulant was pressed tightly to adhere to the active area of device with a pressure of 30 kPa for 3 min while excluding the excess bubbles. The EP encapsulant can be solidified after being placed at room temperature for 5 min. The thickness of EP layer was 84.67 μm. This encapsulation procedure was carried out in ambient air. The POE encapsulation was hot pressed under vacuum condition (50 kPa, 140°C, 10 min) (15), and the thickness of POE layer was 137.26 μm. The epoxy resin used as an edge sealant for the device stability test. Flexible devices were encapsulated with EP, and the structure of the encapsulated flexible devices is PET foil/EP encapsulant/fPSCs/EP encapsulant/PET foil. Specifically, first, the EP encapsulant was heated to 100°C and uniformly applied to PET foil (50 μm); then, the EP encapsulant–coated PET foil was laminated onto the flexible device (5 min, 20 kPa, 100°C) (54, 55). The operation of POE encapsulated flexible devices is similar to that of EP (10 min, 20 kPa, 120°C).

Materials and film characterizations

NMR spectra were conducted by Bruker AVANCE III HD 600 MHz, using DMSO-d₆ as locking solvent (56). FTIR spectra were measured by JASCO FTIR-6100. MALDI-TOF mass spectrometry analysis was performed on a Bruker ultraflex mass spectrometer with a 355-nm SmartBeam laser (200 Hz, 50-μm focus) and 19-kV plate offset voltage (57–59). Data were processed via DateAnalysis 4.0 (Bruker Daltonics). The sample was prepared by mixing equal volumes of dithianol [10 mg/ml in tetrahydrofuran (THF)] and EP (5 mg/ml), and 1 μl was spotted on a stainless steel target, dried at room temperature, and analyzed in positive linear mode. GPC was performed using HPLC-1525 (Waters), and polystyrene was used as a standard sample for GPC calibration. THF was adopted as an eluent (60, 61). The DSC and thermogravimetric analysis (TGA) were conducted using a METTLER TOLEDO instrument (TGA/DSC3+). The optical images were observed by an Olympus BX51 microscope under visible light. TEM images and EDS element mapping images were acquired by the TEM (Talos-F200X) with a bias of 200 kV. For the preparation of TEM samples, the EP polymer was dissolved in dimethyl sulfoxide solvent. The solution was then dropped onto a copper mesh, and the solvent was evaporated for TEM analysis. WAXS measurements were performed by a SAXSpoint 2.0 (Anton paar, Austria, Cu-Kα, λ = 0.154 nm). The single-lap tensile shear strength test was carried out by tensile testing equipment (Instron 5942) with a 500 N load cell. Light transmittance and UV-visible absorption were acquired through a spectrophotometer (Lambda 35, PerkinElmer). The absorption spectroscopy images of perovskite films before and after encapsulation were performed by hyperspectral imager. Photoluminescence (PL) spectra were performed using the FLS980 spectrometer (Edinburgh).

Device characterization

J-V curves of PSCs were conducted by a Keithley 2420 source meter under AM 1.5G irradiation at 100 mW/cm² with a Newport 94023 A solar simulator. A standard silicon cell was adopted to calibrate the intensity of simulated light. The scan speed was set as 0.1 V/s from −0.1 to 1.2 V. Infrared thermal images of encapsulated PSCs were measured by an infrared camera (Fluke Ti300). The encapsulated devices were placed in high-low temperature-humidity test chambers (HS-408 L, 85° ± 0.5°C and 85 ± 0.5% RH) for damp heat stability. The encapsulated devices were placed in a chamber with the temperature cycling between −40° and 85°C. During the temperature cycling process, the rate of temperature change was set to be no more than 100°C/hour, and the temperature should remain stable for at least 10 min at the high and low limits during the testing period. For device stability measurements, on the basis of blanket encapsulation of the PSCs with EP or POE, epoxy resin was used as an edge encapsulant around the device. Meanwhile, to verify the performance of EP self-healing performance on the device stability, epoxy resin was not used on one side of the device to facilitate artificial damage to EP and POE encapsulants. Specifically, we carefully cut the encapsulant on one side along the edge of the devices with a scalpel and ensured that the depths of the scalpel inserted into the encapsulant were the same, aiming to compare the self-healing ability of EP and POE within the same standards. It should be noted that in the thermal cycling stability test, the EP encapsulant was artificially damaged when the temperature rose to 85°C. The operational stability of the encapsulated devices was evaluated by maximum power point (MPP) tracking (YH-VMPP-16) under continuous one-sun illumination (100 mW/cm², white LED) in ambient air at 55 ± 5°C.

Water vapor transmission rate characterization

Al electrodes with a thickness of 100 nm were evaporated on a quartz glass substrate (15 mm by 15 mm). Then, a Ca film with a length of 5 mm, a width of 2 mm, and a thickness of 100 nm was evaporated by thermal evaporation between the aluminum electrodes. Subsequently, the EP with a thickness of 1.2 mm was used to encapsulate Ca film. After that, the encapsulated devices were placed in an environment of 20°C and 90% RH. The insulating calcium hydroxide produced when water vapor infiltrated into the Ca film can cause a change in the conductivity of the Ca film. Therefore, a digital source meter (Keithley 2400) was used to continuously measure the changes in device resistance values and obtain the curves of conductance values over time. Last, WVTR can be calculated according to the following Eq. 1 (62)

$$\mathit{WVTR}=-n\left(\frac{M_{{\mathrm{H}}_2\mathrm{O}}}{M_{\text{Ca}}}\right)\mathrm{\delta }\mathrm{\rho }\frac{\mathrm{l}}{w}\frac{\mathrm{d}\left(\frac{1}{R}\right)}{\text{d}t}$$ (1)

where n is the molar ratio of chemical reaction, and its value is 2. $M_{{\mathrm{H}}_2\mathrm{O}}$ and $M_{\text{Ca}}$ are the molecular weight of H₂O and Ca, respectively, $\mathrm{\delta }$ is the resistivity of Ca (3.91 × 10⁻⁸ ohm·m), $\mathrm{\rho }$ is the density of Ca (1.55 × 10⁶ g/m³), l and w are the length and width of Ca film, and R is the resistance of Ca.

Lead leakage characterization

The amount of lead leakage was analyzed by ICP-MS (PerkinElmer NexION 350D). The standard curve was acquired by configuring standard lead solutions with different concentrations (0, 1, 10, 25, 50, and 100 ppm). The lead SQE was calculated by $\text{SQE}=(1-\frac{\text{leaked lead from ecapsulated perovskite}}{\text{leaked lead from pristine perovskite}})\ast 100\%$ (5).

Computational method

The molecular models were built using the BIOVIA Materials Studio (Dassault Systèmes BIOVIA) (63). The molecular dynamics simulation calculations of the interaction energy were performed using the COMPASS III force field from the FORCITE module (64, 65). The Ewald method and the atom-based method were used to analyze the Coulomb interactions and the van der Waals interactions between different components. To simulate the interface adsorption of the EP and substrates (glass and PET), the EP molecular model was constructed, contains 17 segments and each segment consists of six repeated monomers. Then, additional 17 repeated substrate monomers were added into the above molecular models, with a 50-Å vacuum layer added between the two layers.

The geometry optimization between different components was performed using a smart method with an energy convergence criterion of 2.0 × 10⁻⁵ kcal/mol, and force convergence criteria of 10⁻³ kcal/mol Å were used to get a global minimum energy configuration. Then, an equilibrate process was performed under canonical ensemble [(NVT), where N represents the number of particles, V represents the volume, and T represents the temperature] at 298 K for 1 fs. During the simulation, Nose-Hoover thermostat was applied in the temperature control.

Supplementary Materials

This PDF file includes:

Supplementary Text

Notes S1 and S2

Figs. S1 to S37

Tables S1 to S6